2 papers
cs.OS2026
AutoUVM: Automated Prefetching Framework for LLMs under UVM Oversubscription
Mao Lin, Hui Feng, Xianzhong Ding +3
Large language models (LLMs) increasingly exceed the memory capacity of commodity GPUs, making memory oversubscription common in practical deployments. NVIDIA Unified Virtual Memor…
cs.PF2026
HybridGen: Efficient LLM Generative Inference via CPU-GPU Hybrid Computing
Mao Lin, Xi Wang, Guilherme Cox +2
As modern LLMs support thousands to millions of tokens, KV caches grow to hundreds of gigabytes, stressing memory capacity and bandwidth. Existing solutions, such as KV cache pruni…